There seem to be concerns in the academic community that analytics used to predict human behavior may raise ethical issues. How far can a society rely and act upon a prediction that may turn out to be false? Should the trial and error approach be applicable at all costs?
Makes me think of the film "Minority Report". Maybe you'll find inspiration there. :)
Your question is a bit vague, could you elaborate? Do you mean ethical problems with trusting predictions, or with making such predictions?
I also don't understand the trial and error comment? Usually such algorithms are tested pretty thoroughly before deployment.
Check out :
SADIN, Éric (2011). La société de l’anticipation, Paris : éditions inculte. pp. 77-119. (in French)
KITCHIN, Rob (2014). "Thinking Critically about and Researching Algorithms", The Programmable City Working Paper 5, pp 1-29.
ANANNY, Mike (2016). "Towards an Ethics of Algorithms: Convening, Obervation, Probability, and Timeliness", Science, Technologies, & Human Values, 2016. Vol. 41(1), pp 93-110.
Herman,
Yes, I mean ethical problems with trusting predictions.
The trial and error comment refers to applying predictions to real life situations, not the testing stage of algorithms.
I'll speak as an engineer working with such algorithms rather than as an ethicist. who would probably have more profound insights.
Often the approach is that the algorithms do not need to be perfect for us to use them. They must just be better than what we're doing at the moment (maybe human predictions? Or nothing?) You could always flip the question around to ask, "if I have a tool with a lower error rate than my judgement, or a lower cost than doing nothing, and I don't use it, am I not culpable?" I guess it is a very utilitarian approach (of which I am generally a bit sceptical), but I think one should weigh the cost of using it against the cost of the alternative. The costs could be incommensurable or unquantifiable as in Minority Report, but that would be context-specific. The grounds for deploying an algorithm with unquantified effects/costs is dubious, and then the arguments would be comparable to using untested medical interventions, I guess.
I'm still not quite sure what you mean by the trial and error approach. If it has been tested well before deployment, there should be no trial and error. Are referring to live human trials e.g. facebook testing emotional manipulation/prediction with news feed sentiment changes?
I'll give you an example to think your question.In the area of genetic testing, first,not all predicted diseases could not be tested;second,even if one is able to know the disease,but there is no suitable method to cure it.which likes that we are live in the world with risk, we still live a happly life in most time. so the key ethical questions is 1.for whoes interest? whether the less people should be ignored to get the most people most interest?2.we have the right to know something,whether means that we have the right to not know something?3.what ls freedom?
- Badges
- Science topic
- Similar topics
- Philosophy
- Ethics